The invention relates to a method for producing a three-dimensional (3D) digital x-ray image by using digital x-ray images taken of the object from at least two different directions as input for a reconstruction algorithm. A reconstruction algorithm refers to an algorithm by means of which two-dimensional or three-dimensional x-ray image information on the object being imaged is reconstructed, that is, produced.
Conventionally, the input used for a reconstruction algorithm in 3D x-ray imaging consists of a selection of projections taken from different directions, giving as a result a three-dimensional voxel representation of the tissue being imaged. For example, U.S. Pat. No. 6,611,575 B1 (Assignee General Electric Company) describes three-dimensional imaging using projections taken from different directions as a starting point, the projections being entered in the reconstruction algorithm in the form of spaced-apart planar images. These planar Images are then used as input for the chosen 3D imaging method. The imaging method may be, for example, Volume Rendering or Surface Rendering or a combination of the two.
It is often necessary to select a region of interest (ROI) within the object and to image each projection so that only the ROI is shown in the image. Reasons for this are the minimization of the radiation dosage and the size limitations of the detector. This type of 3D imaging is called local tomography. The problem with local tomography is that objects at a distance from the ROI can be seen in the projections and cause artifacts (errors due to calculation) in the reconstruction.
A technique long since known by dentists and based on simple local tomography is tomosynthetic imaging, in which a two-dimensional slice inside a tissue is imaged accurately and areas at a distance from the slice are shown unclearly. Tomosynthetic slice images can be produced by any of the following methods:    a) By moving the film/detector and radiation source during imaging, which blurs objects at a distance from the in-focus plane and images objects on the in-focus plane sharply.    b) By taking a series of static images, which are appropriately moved and summed to produce slice images corresponding approximately to those produced under a).    c) By combining a) and b): a series of slightly blurred slice images complying with a) are taken, which are then moved and summed according to b).